Predicting the failure of two-dimensional silica glasses

TL;DR

This paper uses machine learning to predict failure in 2D silica glasses from their initial structure, providing interpretable insights into failure mechanisms and demonstrating transferability to different samples and experimental data.

Contribution

It introduces an approach combining deep learning and Grad-CAM for failure prediction and interpretation in disordered solids, applicable to experimental images.

Findings

Predictions can be transferred across different sample shapes and sizes.

Grad-CAM maps relate to topological defects and local energies.

Machine learning provides interpretable failure predictions.

Abstract

Being able to predict the failure of materials based on structural information is a fundamental issue with enormous practical and industrial relevance for the monitoring of devices and components. Thanks to recent advances in deep learning, accurate failure predictions are becoming possible even for strongly disordered solids, but the sheer number of parameters used in the process renders a physical interpretation of the results impossible. Here we address this issue and use machine learning methods to predict the failure of simulated two dimensional silica glasses from their initial undeformed structure. We then exploit Gradient-weighted Class Activation Mapping (Grad-CAM) to build attention maps associated with the predictions, and we demonstrate that these maps are amenable to physical interpretation in terms of topological defects and local potential energies. We show that our predictions can be transferred to samples with different shape or size than those used in training, as well as to experimental images. Our strategy illustrates how artificial neural networks trained with numerical simulation results can provide interpretable predictions of the behavior of experimentally measured structures.